For readers who have experience with survey administration in Africa, Crystal Biruk's Cooking Data is a nostalgic trip. Not the kind of nostalgia you get when an old favourite from high school comes on the radio and you jump to turn up the volume though. It is more like the kind of nostalgia you get when you find one of your undergraduate research papers in a forgotten box. That paper meets the minimum threshold for being a research paper and if you are not too critical you will admit to yourself it is not all that horrible for what it is. But the older and wiser you recognises you could do much better now and your earlier work is a little embarrassing in the present light. Biruk's text brings up that kind of nostalgia. The retrospection hurts but you are a better person for recognising the things that passed unnoticed and/or unexamined years ago.

Cooking Data follows Biruk's work as an anthropologist on a number of demography projects in Malawi, most notably the Longitudinal Study of AIDS in Malawi (LSAM) and the Marriage and Youth Project (MAYP). The demographers working with Biruk desire ‘high-quality, clean data’ and dream of ‘data production on a well-oiled assembly line’ (p. 24). Biruk herself was both an observer and participant alongside demographers in survey design, data collection and dissemination of results. For this project, however, she puts on her ethnography lenses to engage in ‘slow research’ and ‘granular analysis’ not primarily of the data collected but the process that created the data and the information we can glean from the power dynamics baked therein (p. 28).

Bookended by a thoughtful introduction and conclusion, Cooking Data has five empirical chapters. Chapter 1 focuses on survey design and takes readers from the asymmetric relationship between proposal writers and their in-country collaborators through the challenges of translating concepts like wealth reliably into numerous local languages. Chapter 2 considers the local enumerator who often jumps from project to project looking to eke out part of a living while adjusting survey delivery and even question wording to make the project more context appropriate and/or increase the rate of pay by speeding up the often piecemeal work. Chapter 3 uses LSAM's provision of two bars of soap to respondents to explore questions of informed consent, the value of data, and feigned cultural authenticity as a way to build rapport. Chapter 4 looks at the way dozens of judgement calls and improvisations by enumerators out in the field can be made to appear like standardised operations on a data assembly line. Chapter 5 follows data from the field as it metamorphoses into evidence in research journals and policy discussions where the stories told by hundreds of Malawians are converted into culturally biased discussions many cannot read or understand in an attempt to weigh in on how to improve their situation.

I recommend this book to anyone who does survey work in Africa. You will recognise yourself in it and maybe find a more empathetic way to collect data the next time you head out. This book would be an excellent gift from anyone who collects primary data to give her/his favourite colleague who exclusively uses canned data sets to conduct analysis. There is so much effort that goes into ‘clean’ data and so many decisions underlying the numbers that are completely out of even their peripheral view. This book could be very eye opening. In my university's research methods sequence I plan to pair the text with a more traditional textbook that covers the how-tos of survey data collection. Its prose is scholarly but accessible and Biruk does a good job of marrying theoretical concepts to real world examples. Cooking Data will provide something useful for students who have learned how to collect data using best practices in a completely antiseptic environment. Thanks to Biruk's text, I now have a valuable ally in convincing them that best practices always make way for pretty good, and sometimes less than pretty good, practices in the messy and unpredictable real world. The nature of these pretty good practices shapes the characteristics of the data we are trained like good positivist social scientists to look at as ‘natural’ and ‘clean’. Hopefully my students will come to appreciate amazing data collection while simultaneously being cognisant of the relationships and power dynamics that shape the numbers and categories and ultimately our analysis of them.